Error correction practices of Polish and American teachers. (Linguistics).
If a layperson were to be asked to enumerate the most distinguishing features of interaction in the foreign language classroom, he would probably insist on placing the provision of corrective feedback at the top of his list alongside direct instruction, teachers' questions, directives and explanations. The place of error correction in instructed second language acquisition is a highly controversial issue, however, and the attitude to errors committed by learners has undergone considerable modifications in the last few decades.
At the time when audioligualism was in its heyday, teachers, educators and researchers went to great lengths to prevent learners from making mistakes as it was believed that otherwise erroneous habits would be formed and a lot of time would have to be spent attempting to eradicate them. (1) Since, at that time, errors were generally considered to be the outcome of learners' mother tongue interference, Contrastive Analysis was employed to pinpoint the similarities and differences between pairs of languages and, thus, help teachers predict and focus on areas of potential difficulty. The situation changed dramatically with the advent of Error Analysis thanks to which it was discovered that the majority of errors made by second language learners were intralingual in origin rather than related to first language interference (cf. Ellis 1985), as well as the development of interlanguage theory with its claim that learner language was a system in its own right and that second language acquisition was the result of th e process of hypothesis formation and testing (cf. Corder 1967; Selinker 1972). Such views cast doubt on the notion of error itself as inaccuracies in learners' linguistic production could be interpreted as indispensable for language development, and it could be claimed that although some forms in their output did deviate from the native-speaker norm, learners still behaved grammatically in the sense that they drew on the rules they had internalized. The changing views on the nature of language development soon brought about significant modifications in second language pedagogy, which is best exemplified by the appearance of communicative language teaching with its focus on meaning and fluency rather than form and accuracy. Consequently, correcting learners' errors began to be looked upon with suspicion and some researchers went as far as claiming that it might even be detrimental.
However, as is the case with such controversial issues concerning language instruction as teaching the formal aspects of the language code or using the students' mother tongue in the classroom, adopting extreme positions of the kind discussed above does not appear to contribute anything of substance to second language pedagogy. Quite a few researchers currently agree that communicative language teaching in its pure form may not be sufficient for the development of full-fledged communicative competence in the target language and argue that, ideally, second language classrooms should create opportunities for students to participate in both meaning- and form-focused instruction (cf. van Lier 1991; Ellis 1992; Johnson 1996). If such a line of reasoning is adopted, skillful and timely provision of corrective feedback can be viewed as an important tool assisting learners in testing their hypotheses about the target language and, thus, moving along the interlanguage continuum. It also becomes fully warranted to inve stigate the nature of teachers' error correction practices, the impact of these practices on students' output and, most importantly, their contribution to the development of target language proficiency.
This paper attempts to build on the reasoning presented above by discussing the influence of such variables as teachers' linguistic background and teaching experience on the provision of corrective feedback as well as assessing the potential effect of that feedback on the development of the students' communicative competence in the foreign language context.
2. Some relevant theoretical issues and research findings
In any communicative exchange speakers derive from their interlocutors information on the reception and comprehension of their messages which is generally referred to as feedback or negative evidence. While in naturalistic contexts feedback is the outcome of negotiation between speakers and listeners, in the classroom it is mainly provided by the teacher as a result of his superior knowledge as well as the communicative asymmetry inherent in any instructional setting.
As can be seen from Figure 1, negative evidence in second language acquisition can be either preemptive, where the teacher provides students with some kind of formal instruction before an error occurs, or reactive, which involves informing the student that his utterance has not been comprehended or is faulty in some way. One course of action that the listener can take when providing reactive negative feedback is to implicitly indicate the trouble source and, thus, initiate a repair sequence. This can be accomplished by means of recasting (i.e. repeating the speaker's original utterance simultaneously providing the correct form) as well as repetitions, confirmation checks and clarification requests, which are not included in the above model, but are extremely important in that they might lead to negotiation of form and meaning and, similarly to recasts, enable the learner to "notice the gap" in their linguistic knowledge (cf Schmidt and Frota 1986; Schmidt 1990, 1992). The provision of implicit negative eviden ce can further result in the speaker modifying his original utterance or the occurrence of a communication breakdown. The focus of the study reported in this paper, however, will mostly be on situations in which the listener overtly corrects some aspect of the learner's output, or, to put it differently, on "a reactive second move of an adjacency pair to a first speaker's or writer's utterance by someone who has made the judgement that all or part of the utterance is linguistically or factually wrong" (James 1998: 236-7).
As has already been pointed out, there is no general consensus among researchers as to the value of overt error correction in second language acquisition. The adherents of communicative and natural approaches to language teaching claim that providing learners with explicit negative evidence is, at best, ineffectual and, at worst, can constitute an obstacle to second language development. Krashen (1982), for instance, argues that error correction puts learners on the defensive and, as a result, they tend to avoid using difficult structures and focus on form rather than on meaning, all of which is detrimental to acquisition. He concludes that "even under the best of conditions, with the most learning-oriented students, teacher corrections will not produce results that will live up to the expectations of many instructors" (1982: 119). Another argument against the usefulness of negative feedback comes from studies into first language acquisition which suggest that caretakers usually emphasize factual content and effectiveness over form in interactions with children acquiring their mother tongue (Allwright and Bailey 1991: 86). Similarly, research into repair in native language conversations has shown a strong preference for self-initiated self-completed repair among speakers of English as well as a virtual lack of other-initiated other-completed repair (cf. Schegloff et al. 1977; Gaskill 1980). (2) Finally, many researchers believe that in acquiring a second language learners pass through a number of developmental stages and, therefore, error treatment involving grammatical structures which are not part of the students' interlanguage is bound to be ineffective (cf. Corder 1967; Pienemann 1984, 1989).
Although they acknowledge the dangers connected with excessive correction in the language classroom, other researchers point out that providing learners with explicit negative evidence can sometimes play an important role in the process of instructed second language acquisition. It has been argued, for example, that since learners' output serves as input for both the speaker and other students, lack of negative evidence in the form of corrective feedback can serve as an affirmation of their incorrect hypotheses about the rules for generating the target language (cf. Schachter 1988). Another argument in favor of error correction is provided by White (1987, 1991), who claims that some grammatical contrasts between the first and second language are unlearnable from positive evidence alone and, therefore, both preemptive and reactive negative evidence is sometimes indispensable to further acquisition (cf. Long and Robinson 1998). Some researchers also point out that bringing errors to the learners' attention help s learning as it contributes to the process of consciousness-raising, which has been hypothesized to play an important role in second language acquisition (cf. Edmondson 1985; Lightbown and Spada 1990; Lightbown 1992; Sharwood-Smith 1980, 1991,1993). They claim that an interpretation of communicative language teaching that excludes form-focused activities and error correction may in fact make it difficult for learners to obtain quality input and, as research into French immersion programs has shown (Harley and Swain 1978; Swain 1985, 1989), does not necessarily lead to grammatical accuracy. As Lightbown and Spada (1990: 443) put it: "accuracy, fluency and overall communicative skills are probably best developed through instruction that is primarily meaning-based but in which guidance is provided through timely form-focused activities and correction in context".
The two divergent theoretical positions have brought about a number of studies investigating the effect of error correction on second language development, but, as is the case with many other facets of classroom interaction, the findings of this research are mixed and inconclusive. One of the studies that failed to find a positive relationship was conducted by Chaudron (1986) who reported that only 39 per cent of errors treated by the teacher in French immersion classes were eliminated in the next student utterance. Similar results were obtained by Brock, Crookes, Day and Long (1986) who admitted, however, that the absence of observable effects of corrective feedback in the short term does not mean that they do not exist over time as "a learner may require a certain amount of time to make use of negative input, and in the interim will continue to operate with old, as-yet-unmodified hypotheses" (1986: 235-36). On the other hand, Ramirez and Stromquist (1979) found a positive correlation between the correction of grammatical errors and general gains in linguistic proficiency, but this study did not investigate the relationship between the correction of specific errors and the subsequent elimination of these errors in the learners' interlanguages. Limited support for the role of corrective feedback is also provided by Carroll, Roberge and Swain (1992) who observed that corrective feedback helped learners memorize those individual words on which treatment was administered, but failed to assist them in restructuring their grammars and generalizing the corrections to novel forms. Finally, Lightbown (1992) compared the effects of immediate corrective feedback in communicative and practice activities, and found that although learners were able to self-correct in both conditions, it was only in the former case that the self-corrections became intake, as evidenced by the use of the targeted forms outside the classroom. (3)
Apart from attempting to determine the role of explicit negative evidence in second language development, researchers have also thoroughly examined the decisions teachers have to make when they react to learners' oral errors. As Figure 2 illustrates, error treatment is an extremely complex process and requires a number of quick choices that the teacher has to make before he comes up with a reactive move to the student's erroneous response. First, the error has to be noticed, which is particularly relevant in the case of non-native teachers whose target language competence may be lacking in some respects. If an error is in fact noticed, it has to be decided whether or not it should be subject to some kind of treatment. As research suggests (Nystrom 1983; Chaudron 1986), there is considerable variation among teachers with regard to the extent to which they provide explicit negative evidence as well as the types of errors they are most likely to correct. Depending on the theoretical approach they favor as well a s their teaching styles, some teachers tend to correct any utterance that deviates from the target language norm, whereas others choose to ignore linguistic inaccuracies and focus exclusively on content errors. In addition, it has been found that discourse, content and lexical errors generally receive more attention than phonological and grammatical ones. As Long (1977) points out, however, teachers are extremely inconsistent when it comes to providing feedback since they correct errors committed by some students and not by others, or those produced in one part of a lesson but not in another. It has also been suggested that, at times, teachers in fact induce errors on the part of the learners through forming their questions or explanations in a particular way (cf. Edmondson 1986).
Having decided to treat a particular error, the teacher may choose to deal with it immediately, delay the treatment until the learner finishes delivering the message he is trying to convey, or postpone the provision of negative feedback for a longer period of time. It has been argued that immediate error correction is disruptive and provides negative affective feedback, which can make learners reluctant to speak. (4) On the other hand, however, there is a danger that feedback provided a considerable amount of time after the occurrence of the deviant form in the learners' output can be less effective (cf Long 1977). No matter when the treatment takes place, the teacher also has to decide how to react to the given error. He might, for instance, rephrase the incorrect response or use repetitions in order to inform the learner that an error has been committed, as well as to provide him with some information on its location and identity. Research has shown that teachers often give more than one type of feedback si multaneously, that many of their reactive moves go unnoticed by the students, and that the same reaction can be used for different purposes (cf Long 1977). This state of affairs is best summarized by Nystrom who writes: "the fact remains that each interactive situation is quite complex, and teachers typically are unable to sort through the feedback options available to them and arrive at the most appropriate response" (1983: 170). (5)
The final decision that the teacher has to take pertains to who should perform the correction and has been of particular interest to researchers. Since it is only the learners who are actually capable of making permanent changes to their interlanguage systems, it has been generally assumed that self-correction is the most beneficial way of repairing communication breakdowns. Researchers have found (e.g., van Lier 1988) that, in contrast to general conversation, where self-initiated self-repair is prevalent, the second language classroom is usually characterized by a predominance of repair initiated and completed by the teacher. This does not mean, however, that other types of repair cannot occur as the type of explicit negative evidence in an instructional setting can be conditioned by the context jointly created by the teacher and the learners, as well as the individual preferences of the teacher and the pedagogic purpose of a given task. Kasper (1986) reports, for instance, that in the language oriented pha se of an English lesson in Grade 10 of a Danish gymnasium, troublesources were identified by the teacher but the repair work was done either by the learner who committed the mistake or by one of his peers. In the contentcentered phase of the same lesson, on the other hand, there were many cases of self-initiated self-completed repair and there was a tendency to avoid interruption of the ongoing discourse.
To make matters even more complicated, the incidence and character of error treatment also depend on whether the teacher is a native speaker of the target language. The general finding has been that native speakers are much more tolerant of errors than non-native speaking teachers (Hughes and Lascaratou 1982; Davies 1983; Sheorey 1986). This situation may be due to the fact that, while the former are more concerned with the effect of a given error on comprehensibility, the latter are usually more influenced by their ideas of what constitutes the rules of the target language (Hughes and Lascaratou 1982). (6) The corollary of these disparate approaches is that native speakers tend to judge lexical errors as more serious than grammatical ones and have been found to correct global grammatical errors as more likely to interfere with comprehension than local ones (Tomiyana 1980; Khalil 1985). (7) In contrast, non-native speakers are much more likely to provide corrective feedback on morphological and functor errors , and, at the same time, attach relatively little importance to errors affecting overall sentence organization as well as those involving the inappropriate use of lexical items.
As can be seen from the above overview, error correction in the language classroom is an extremely complex process and no recommendations can be made with absolute certainty. What the studies reported above seem to suggest, however, is that focus on form through explicit negative evidence has an important role to play in second language development, and that its provision will be more effective if the teacher bases his decisions regarding correction on a careful analysis of the teaching context. Some principles aimed at making this decision-making process more efficient are proposed by McPherson (1992) who investigated error correction from the perspective of the leamer. She argues that teachers should use correction methods which are compatible with the leamers' current language proficiency, encourage purposeful learner involvement by allowing opportunities to self-correct or analyze the errors, and provide corrective information or clarification in a positive manner. Presumably, it is by following these gui delines that negative feedback can facilitate rather than hinder language learning.
In the study to be reported in the following pages we will follow Chaudron (1986: 67) in defining errors as linguistic forms or contents that differ from native speaker norms or facts as well as any other behavior signaled by the teacher as needing improvement. When investigating errors occurring during lessons analyzed for the purposes of this research project, we will adopt a simple taxonomy broadly distinguishing between content, discourse, grammatical, lexical and phonological errors, which was also used by Chaudron (1986) in his study.
3 Subjects, data collection procedures and methods of analysis
The investigation of repair presented in this paper was part of a large-scale research project designed to explore interactive processes during regularly scheduled lessons conducted by ten Polish and ten American teachers in secondary schools across the country. The study being of a descriptive-exploratory nature, there were no specific criteria on the basis of which schools were selected for data collection and the only two factors that were taken into account were the presence of American teachers as well as their and their Polish counterparts' willingness to participate in the project. All of the American teachers were Peace Corps volunteers whose teaching experience varied from 6 months to 3 years (for 8 of them it was less than 2 years) and most of whom had only completed a three-month Pre-Service Training in TEFL Methodology, described by one of them as "swift and of little help". (8) Despite the communicative focus of the training, only two volunteers reported close adherence to that teaching method, w hile another four stated that they subscribed to the communicative approach "loosely", "whenever possible" or that they "did not strictly follow that method". The remaining four volunteers claimed that they used a variety of teaching methods depending on the class they taught, and one of them even characterized herself as an adherent of the Total Physical Response Method. As for the Polish participants, with one exception all of them were frilly qualified teachers with university degrees in English and teaching experience at the secondary school level varying from 2 to 20 years. Since only two of them had taught English for less than five years prior to the experiment, the group as a whole can safely be described as having considerable teaching experience, which contrasts sharply with the Peace Corps volunteers participating in the project. (9) As regards their methodological preferences, 6 Polish participants stated that they did not adhere to any particular teaching method and the other four reported using "a mixture of many approaches", and admitted that the choices they made depended on class level and many other factors.
A total of 842 students participated in the lessons recorded for the purposes of this project. At the outset, it was the intention of the researcher to focus exclusively on third grade students in order to make comparisons more reliable, but it proved to be unfeasible due to the reluctance of many Polish teachers to participate in the project as well as the fact that in many schools both Polish and American teachers did not have classes at this level. Consequently, it was decided not to exclude any level from the study and, in retrospect, this decision appears to have been fully justified as it soon turned out that the students' level of target language development had little to do with class level and most of the classes observed could aptly be described as mixed-ability groups.
Two instruments of data collection were used for the purposes of this study with an eye to getting multiple perspectives of the phenomenon under investigation and thus examining it in a comprehensive and objective manner. They included audio recordings of regularly scheduled lessons and questionnaires distributed among both the teachers and some of their students. The choice of the instruments as well as their relatively unobtrusive application were intended to minimize intervention in classroom proceedings, and thereby capture interactive processes as they naturally occurred in the classroom. Such an approach reflects the belief on the part of the researcher that "experimental research coupled with statistical tests of significance have so far failed in making progress" (van Lier 1988: 3), and that it is ethnography, with its focus on a thorough description of the investigated phenomena, that is most appropriate for conducting classroom-oriented research.
The audio recordings were made in different secondary schools across the country and they comprised three complete lessons conducted by each of the twenty participating teachers. Prior to each lesson to be taped, the teachers were told that regular classroom proceedings were of interest to the researcher and that they should not abandon their original lesson plan because of the recording. The students knew that they were being recorded, but to reduce anxiety and reactivity, they were not required to hold a microphone or to deliberately speak in its direction. In order to get more reliable and accurate transcripts of classroom interaction, and to minimize the possibility of technical failures, two dicta-phones equipped with condenser microphones were placed in two different locations in the classroom where a particular lesson was taking place.
As for the questionnaires, they were distributed among the teachers and some of the students participating in the study. (10) It was hoped that this introspective method of data collection would provide the researcher with information about phenomena that might not be directly observable in the classroom and, at the same time, enable him to evaluate the extent to which the data gathered by means of audio recordings reflected the teachers' regular classroom practices. Approximately half of the questions in the questionnaire administered to the teachers were open-ended, which was based on a conviction that they were qualified enough to make relevant observations concerning their classroom practices as well as their students' performance without being prompted. Despite some problems with getting the respondents to return the questionnaires in time and the consequent necessity to contact some of them several times, all the teacher questionnaires were eventually collected thus making comparisons easier. Although t he focus was similar, the student questionnaires differed from those handed out to the teachers in that they mostly consisted of highly specific questions with a limited set of answers so as to avoid confusion on the part of the respondents and, thus, reduce the incidence of irrelevant responses. A total of 622 questionnaires were distributed among students out of which 547 were eventually returned thus putting the return rate at 88%.
The sixty lessons recorded for the purposes of this study were later transcribed and, together with the questionnaires, served as a basis for further analysis, which took the form of a combination of the qualitative and quantitative approaches in accordance with the recently widely held view that "we cannot limit observation to what can be measured without ignoring most of the areas that teachers and learners are interested in [and that] it is much more important to break down our questions into those parts for which objective and measurable categories are appropriate, and those for which such categories cannot be neatly devised" (Brumfit and Mitchell 1990: 13). (11) Quantitative analysis was conducted by counting the total number of errors committed by students during a given lesson and the total number of corrections made by the teacher. (12) The corrections were also coded separately depending on the type of error using Chaudron's (1986) taxonomy as a point of reference. (13) The instances of corrective fe edback were also counted with regard to the source of repair initiation and execution, and therefore the numbers of other-initiated other-completed, other-initiated self-completed and self-initiated other-completed instances of repair were also counted. In all these cases, quantitative analysis was conducted by means of simple numerical statistics, which appears to be quite adequate for the purposes of a descriptive-exploratory study, where no specific hypotheses are to be tested. Qualitative analysis involved describing the factors determining the initiation of corrective behavior on the part of the teacher, the ways in which he informed the students that an error had been committed as well as the effect of correction on the students' subsequent production. Some attention was also paid to the context in which explicit negative feedback was provided.
4. Results of the study
One of the most striking differences between the Polish and American teachers participating in this study was that the native speakers turned out to be much more tolerant of errors than their Polish counterparts, which is in accordance with the results of the studies discussed in one of the previous sections (Hughes and Lascaratou 1982; Davies 1983; Sheorey 1986). The analysis of the transcribed data revealed that while the Poles corrected 48.6% of the errors committed by their learners, the Peace Corps volunteers chose to provide corrective feedback on only 30.0% of the deviant forms in the students' speech. As could be expected, there was a lot of individual variation in both groups, with some Poles correcting as few as 30% of the errors and some Americans treating as many as 90%, which might indicate that the teachers' linguistic background and teaching experience were not the only variables that influenced correction rate. On the whole, the results of this quantitative analysis are consistent with the lea rners' perceptions of their teachers' corrective behavior since 39% of the students claimed that their Polish teachers frequently corrected their mistakes, whereas only 26% expressed a similar opinion of the Peace Corps volunteers. As for the teachers, the Poles proved to be quite accurate when it came to describing their attitudes to error treatment, since seven of them admitted that they often corrected their students' errors and the other three reported doing it from time to time. The situation looked much different for the Americans as, contrary to the findings presented above, half of them stated that they often reacted to their learners' mistakes and the other half claimed that they did it from time to time. One reason for this discrepancy could be the belief on the part of some of the Peace Corps volunteers that correction rate is the measure of the teacher's experience and competence, or simply their inability to objectively evaluate the classroom practices they employed.
In contrast to research findings reported above, the Polish and American teachers did not significantly differ in their preferences for correcting different types of errors, but in some cases the differences were quite unexpected. As can be seen from Table 1, there was some predictable but rather slight variation in the frequency with which both groups corrected lexical and phonological errors, but the most surprising finding was that the Americans treated more grammatical errors than the Poles. This might be the outcome of the fact that, quite unexpectedly, a few American teachers concentrated almost entirely on introducing and practicing grammatical structures, and doing related exercises, whereas only two or three lessons conducted by their Polish counterparts had such a clearly defined structural and activity-oriented focus. As was the case with the overall correction rate, the teachers' perceptions concerning the errors they were most likely to correct were somewhat inconsistent with the above findings. One example of this disparity is that only two American and as many as six Polish teachers reported correcting lexical errors on a regular basis, and, at the same time, all of the Poles and seven Peace Corps volunteers expressed their preference for treating grammatical errors. It should also be noted that half of the Polish and American teachers claimed that they were very likely to react to errors that block communication irrespective of their nature, and that many of them added that whether or not corrective feedback was provided depended to a large extent on the focus of a given activity as well as the phase of a given lesson.
Since, as has been pointed out elsewhere, other-initiated other-repair is extremely rare in general conversation, it was assumed that the largely inexperienced Peace Corps volunteers would be much less likely than their Polish counterparts to correct the students errors in this manner. As shown in Table 2, this assumption proved to be erroneous in the case of the lessons analyzed for the purposes of the study because the American teachers were found to initiate and complete repair sequences by themselves 5.3% more often than the Poles. At the same time, the students self-corrected 2.1% more deviant forms in their output during the lessons conducted by the Poles than by the Americans, which might indicate that it was the former who were somewhat more likely to provide their learners with useful feedback on their production and, thus, to promote self-repair. It is also noteworthy that it was the students instructed by the Polish teachers who more frequently indicated the problems they were experiencing with som e grammatical structures or vocabulary, and were thus more successful in obtaining corrective feedback. Unexpected as they may be, the above differences are rather slight and show that the teacher's linguistic background and teaching experience had little influence on the provision of corrective feedback during the lessons analyzed for the purposes of this project. This finding was confirmed by the results of the student questionnaires, since the learners had similar perceptions concerning the opportunities they had for self-correction irrespective of who provided instruction. As for the teachers' perceptions of their own practices, they were mixed, with most Polish and American teachers reporting that they employed all types of repair depending on the nature of a particular error as well as the context in which it was committed.
Notwithstanding the above findings, the occurrence and scope of error correction were to a large extent determined by the format of a given lesson as well as the nature of the task that the students were requested to perform. Irrespective of their teaching experience and linguistic background, the teachers tended to be considerably more tolerant of errors when the focus of a given lesson or its phase was on a genuine exchange of ideas, and the students were expected to make frequent contributions to the ongoing discourse. In such cases, the provision of corrective feedback was mostly confined to lexical errors as well as errors that blocked communication, while grammatical and phonological problems in the students output received much less attention unless they were extremely serious. At the same time, form-focused and activity-oriented instruction was characterized by a high incidence of error correction, and it was mostly errors involving grammar and phonology that were likely to be treated.
Such a differential approach can perhaps be attributed to the teachers' desire to provide the students with maximum practice opportunities during meaning-focused instruction, and it also indicates that a lack of experience does not affect the amount of correction when the focus is on practicing a grammatical structure or reading out some sentences form the textbook. A more practical reason for such a state of affairs might be that it was much more difficult to notice and correct mistakes when the teachers had to concentrate on the content rather than the formal aspects of the students' contributions in order to be able to follow up on their ideas as well as to control the direction of a given exchange. This task became even more complex when the learners produced lengthy utterances and the nature of their responses was unpredictable due to the local construction of discourse. In such cases, even if a student's response was fraught with deviant forms, it was typically errors occurring at the end of a given seq uence or those that the students themselves attempted to self-correct that were most likely to be discerned and treated by the teacher. Despite such problems, however, a certain amount of corrective feedback within the context of genuine communication is clearly indispensable since, as research suggests, it enhances the quality of input and can sometimes help learners notice and eliminate deviant forms from their developing interlanguage systems (cf Lightbown 1992; Sharwood-Smith 1993; Long and Robinson 1998). Examples of the teachers' corrective feedback within the context of meaning- and form-focused instruction follow:
(1) ((a student talks about the role music plays in her life and even though her response contains a number of different errors, the teacher chooses to correct the one that occurs at the end of the utterance and that the student herself struggles with)
L. (..) and when I'm ... when I'm listening the music, I ... can, I can to say inspiration from this music ... so I can create something ... it's very nice to ... to listening the music when I'm doing something or ... but I think that the music have some ... had ... has something, have something
T. has, has
L. has some influence on my life
(2) ((a student talks about her impressions on hearing a piece of music and the teacher ignores all the errors she makes except a lexical one that causes a communication breakdown)
L. I saw that] a naked woman ... was (riding) the horses and she had long hair .. and they are on ... on the wind, they ... the hair
T. on the what?
L. on the (wings)
L. she was on a horse
L. and ... she had long hair
L. and the hair was ... I don't know
(3) ((students make sentences with will one by one and the teacher corrects many errors involving grammar in general and the target structure in particular)
L1. my mother will ... will be go in the church
L2. to the
T. will go TO church tonight ... mmh Agnieszka
L3. (..) I will have birthday ... my ... I will gratulate my friend tomorrow ... my brother day twentieth July /dzuli/ finished school
T. will finish school ... OK, Ania
When the teachers decided that they should react to some incorrect form in the learners' output, they usually did so immediately either by interrupting the flow of speech or waiting for the student to complete his turn. On a few occasions, however, the teachers chose to delay the treatment until the end of some activity, especially if it involved a genuine exchange of ideas, and then dealt with some more serious or persistent errors in more detail. This sometimes happened upon completion of pair or group work when the teacher wanted all the learners to benefit from his correction, and sometimes led to lengthy grammatical or lexical explanations. Quite surprisingly, it was two Peace Corps volunteers who were observed to devote parts of their lessons to special feedback sessions during which they instructed students to correct their peers' erroneous production that occurred during role plays or discussions. Although such delayed error treatment is perhaps less disruptive and it does not provide the negative aff ective feedback associated with immediate error correction, its effectiveness may be limited since the students' processing mechanisms are more likely to be activated when they receive negative evidence right after they have produced a deviant form (cf Long 1977).
As noted above, it was other-initiated other repair that predominated during the lessons investigated for the purposes of this study, which stands in contrast to what happens in naturalistic settings and is, therefore, frequently considered to have little value for language acquisition. Such a situation, however, is inevitable in the language classroom, where the teacher is usually the most proficient speaker of the target language and it is his responsibility to ultimately resolve communication breakdowns as well as to ensure the availability of high quality input that students can utilize in restructuring their grammars. Even though, for obvious reasons, it would be beneficial to provide students with opportunities to self-correct by repeating their faulty utterances, informing them of the commission of an error, or using clarification requests and confirmations checks, such practices are too time-consuming and too disruptive to be resorted to on a regular basis. Additionally, students frequently failed to successfully correct their faulty utterances in response to the teacher's prompts or questions, and, in the end, it was the teacher or a peer who had to supply the correct form. The indispensability of corrective feedback in an instructional setting as distinct from naturalistic contexts is further evidenced by the fact that the students relatively rarely initiated repair on their own and even when they did so, they were oftentimes unable to fix the problem. It should also be noted that the teacher was not the only source of negative evidence and that peer correction occurred quite frequently during the lessons analyzed for the purposes of this study. This type of corrective feedback commonly resulted from the teacher's general nomination, but there were also instances where the learners self-selected to correct or complete their peers' utterances that they perceived as lacking in some respects. The examples cited below illustrate some of these points:
(4) ((a student self-corrects after the teacher's prompt, but the utterance still contains a lexical error)
L.. they washed the boat
L. they watched the boat ... on the
T. ... on the sea
(5) ((a student modifies his utterance as a result of the teacher's clarification request)
L. we don't have official results /rezults/
T. (...) you don't have what?
L. official charts
(6) ((a student fails to replace the incorrect vocabulary item in response to a clarification request and the teacher fixes the problem)
L. the problem is that in a small town in north of England ... called Blackthorpe ... there is a big international factory Technotronics which employees over [twelve hundred
T. which what?] ... (2) which emp-
(7) ((a students fails to self-correct in response to the teacher's indication that an error has been committed and a peer treats the error of his own accord; the teacher confirms the correction)
L1. so when he was eleven years ... ah ... ah
T. do we say eleven years? (...) when he was eleven?
T. years old or when he was eleven (...)
(8) ((the teacher nominates a student to treat an error committed by her peer)
L1. small, smallger, the smallgest
T. no, listen to ... to ... to ... to the adjective again, well, small ... (3) Magda
L2. small, smaller, the smallest
T. yes, repeat Joan
(9) ((a student notices a mistake and attempts to correct his utterance, but he is unsuccessful and the teacher steps in)
L. when I was in home Malwina ... home Malwina
T. Malwina's home
One measure of the effectiveness of the provision of corrective feedback is the extent to which the teachers' or peers' corrections are likely to be incorporated into the learners' subsequent production. Although there was much variation during the lessons recorded for the purposes of this study, such incorporation appeared to be largely dependent on the intentions of the teacher, the source and type of treatment, the context in which it took place as well as the student's awareness of the incorrectness of the form he produced. In many cases, the teachers did not provide the students with ample time to process the feedback they received, and immediately followed their correction with a question directed at another learner. There were also instances where, presumably for fear of providing negative affective feedback, the teachers did not even indicate that an error had been committed and treated it implicitly by repeating the student's utterance and continuing their elicitation or explanation. On the other han d, the students were more likely to modify their output when the teacher provided them with the correct form only after an attempt at negotiation had been made, or when they had noticed the problem themselves and started the repair sequence of their own volition. Additionally, incorporation of corrective feedback was extremely likely to occur when it was a peer rather than the teacher who corrected a mistake, and when the treatment was provided within the context of a genuine exchange of ideas. Undeniably, such immediate modifications of the students' utterances do not necessarily indicate that the students internalize the form in question, which is reflected in the fact that on some occasions a student repeated the same mistake in a different phase of the lesson.
5. Final remarks and conclusions
As the above discussion demonstrates, there were relatively few significant differences between the Polish and American teachers with regard to providing corrective feedback, the only exception being the differential correction rate for both groups. In fact, the teachers' linguistic background and the length of teaching experience appeared to have only a marginal impact on the nature and focus of repair, which clearly runs counter to some of the available research findings. Apparently, the incidence and character of corrective feedback must have also been considerably influenced by the teachers' beliefs concerning the teaching-learning situation which might reflect, among other things, their experiences as language learners and instructors, their methodological preferences, the training they received as well as personality factors (cf. Richards and Lockhart 1994).
A far more important conclusion to be drawn from the analysis presented in the preceding section appears to be that the process of error correction is so complex, and it involves such a diversity of behaviors on the part of the teachers and their students that its value for the process of instructed language acquisition is extremely difficult to pinpoint. It seems warranted to assume, however, that the provision of a certain amount of corrective feedback is clearly necessary in the language classroom, where communication breakdowns are bound to occur and the incorrect production on the part of some students may serve as input for their peers. This position seems to be particularly valid in the foreign language context, where the learners have little if any exposure to the target language outside the classroom and the hypotheses they form are largely affected by their peers' linguistic production as well as the teacher's reaction to it. It should also be borne in mind that explicit negative evidence plays an i mportant role in bringing the students' attention to the deviant variants in their output and, thus, can significantly contribute to reconstructing their inter language systems. Even though, as a number of researchers argue (e.g., Long 1991; Doughty and Williams 1998; Long and Robinson 1998), it is implicit negative evidence in the form of clarification requests, confirmation checks or recasts that leads learners to self-correct, and is therefore more effective in getting them to notice the gaps in their linguistic knowledge, such a course of action might not always be an option for the teacher. Invaluable as the resulting negotiation of form and meaning seems to be for language acquisition, it is extremely time-consuming and can only infrequently be drawn upon when language instruction is limited to just a few hours a week. Also, as can be seen from the analysis of the data, students oftentimes fail to self-correct when repair is initiated by the teacher and, in such cases, the provision of explicit negative evidence might in fact be the only viable alternative.
Obviously, if these benefits are to be fully exploited, the amount and nature of corrective feedback should be geared to the format of a particular lesson, the task at hand as well as the needs of the learners. This is because the complexity of classroom discourse makes it virtually impossible to come up with a set of guidelines that can be followed in all contexts and circumstances and, therefore, it is for the teacher to decide how much corrective feedback should be provided in a given situation, what errors should be targeted and what form error treatment should take to best assist students in moving along the interlanguage continuum.
Table 1 Corrections of different types of errors as provided by Polish and American teachers Error type Polish teachers American teachers N % N % Content 122 17.9 64 17.3 Phonological 197 28.9 88 23.6 Lexical 87 12.8 70 18.8 Grammatical 218 32.9 128 34.3 Discourse 58 8.5 22 6.0 Total 682 100 372 100 Table 2 Type of repair of learners' errors in lessons conducted by Polish and American teachers Repair type Polish teachers American teachers N % N Self-initiated 31 4.6 5 other-repair Other-initiated 77 11.2 34 self-repair Other-initiated 574 84.2 333 other-repair Total 682 100 372 Repair type American teachers % Self-initiated 1.4 other-repair Other-initiated 9.1 self-repair Other-initiated 89.5 other-repair Total 100
(1.) The distinction introduced by Corder (1967) notwithstanding, the terms "error" and "mistake" are used interchangeably throughout this paper with no theoretical implications intended.
(2.) As Schegloff, Jefferson and Sacks (1977) found, other-initiated self-repair is also common here, but in such cases a confirmation check or a clarification request on the part of the listener is usually preceded by a pause and uttered in a way indicating uncertainty. This is because unmodulated repair-initiation is often associated with disagreement and might constitute a face threat to the original speaker. Schegloff et al. conclude that: "self-correction and other correction are not alternatives. Rather, the organization of repair in conversation provides centrally for self-correction, which can be arrived at by the alternative routes of self-initiation and other initiation -- routes which are themselves so organized as to favor self-initiated self-repair" (1977: 377).
(3.) It may also be relevant that students display an overall positive attitude towards error treatment and usually expect more correction than they are provided with in both communicative and form-focused activities (cf. Cathcart and Olsen 1976; Chenoweth et al. 1983). It was also found, however, that when teachers attempted to take these preferences into account, most students found the ensuing interaction unacceptable (Cathcart and Olsen 1976).
(4.) The distinction between cognitive and affective feedback was made by Vigil and Oller (1976). The former refers to the positive or negative information about the target language forms and the latter concerns the motivational support with which interlocutors provide each other during a conversation. The two types of feedback interact and influence the learners' efforts to attempt revision of their production.
(5.) There is considerable disagreement among researchers over whether this lack of systematicity in providing feedback is desirable. Long (1977), for instance, argues that teachers should be more consistent if they do not wish to confuse their students. Allwright (1975), on the other hand, suggests that inconsistency is inevitable and beneficial as it reflects the efforts on the part of the teachers to cater for the needs of individual students.
(6.) According to Davies (1983), however, the decisions non-native speakers make concerning the provision of explicit corrective feedback are likely to be influenced by a number of factors relating to the particular context in which they operate. Consequently, their judgements will be based on their background knowledge of the syllabus and the textbook used by the learners, as well as explicit knowledge of their mother tongue.
(7.) The distinction between global and local errors was made by Burt and Kiparsky (1972). Global errors can be defined as those affecting overall sentence organization and include wrong word order, missing or wrongly placed sentence connectors, and syntactic overgeneralizations. Local errors, on the other hand, affect single elements in a sentence and include, for example, errors in morphology or morphological functors.
(8.) Three of the ten teachers also claimed to have completed some EFL courses hack in the United States as part of their college education. As one of hem put it, however, they were not "very relevant".
(9.) Obviously, the concept of teaching experience is arbitrary and it could well be argued that it takes much more or much less than 5 years to become an experienced language teacher. Some decision had to be made, however, and what counts here is that there is a considerable difference in this respect between the Polish and American teachers taking part in this study.
(10.) Naturally, since the questionnaires were designed to elicit data on a number of facets of classroom discourse, only a few questions were related to error correction.
(11.) While qualitative analysis was applied to all the sixty lessons recorded and observed in this project, only forty of them were analyzed quantitatively. The main reason for such a decision was the fact that, in many cases, two of the three lessons recorded for one teacher were almost the same except that they were taught to different groups of students, and the outcomes of qualitative analysis suggested that few if any differences were to be expected as far as the occurrence of the phenomenon under investigation was concerned.
(12.) This analysis excluded the errors that the learners themselves detected in their speech, and that they subsequently successfully corrected without any intervention on the part of the teacher or their peers. Such a decision was connected with the fact that self-initiated self-repair is a reflection of the students' monitoring skills rather than the teacher's preferences concerning error correction and, therefore, its inclusion might have somewhat distorted the results of the analysis.
(13.) For the purposes of this study errors were classified as (1) content when a student failed to provide an answer to a question about a text he was supposed to read, provided some information that was known to be incorrect, or failed to provide a form that he had previously been taught, (2) phonological, which included only the instances when the students' rendition of a given segment or stress was totally dissimilar to the target model, and excluded errors involving rhythm and intonation as too intangible, (3) lexical when a student used a word or phrase erroneously, or failed to complete his utterance because he did not recall the requisite vocabulary item, (4) grammatical including both morphological and syntactic errors, and (5) discourse when a student spoke too silently, used his mother tongue in situations where the target language was required or provided an incomplete, but semantically clear response. The classification was not always easy as the distinctions between some error types were oftenti mes extremely subtle and their ultimate classification seemed to be the function of the focus of teacher's corrective feedback rather than the nature of the deviant variant produced by the student.
Allwright, Richard L.
1975 "Problems in the study of the language teacher's treatment of learner error", in: Marina K. Burt -- Heidi C. Dulay (eds.), 96-109.
Allwright, Richard L. -- Kathleen M. Bailey
1991 Focus on the language classroom. Cambridge: CUP.
Brock, Cynthia -- Graham Crookes -- Richard Day -- Michael H. Long
1986 "Differential effects of corrective feedback in native speaker - nonnative speaker conversation", in: Richard Day (ed.), 229-236.
Brown, H. Douglas -- Carlos A. Yorio -- Ruth Crymes (eds,).
1976 On TESOL'77: Teaching and learning English as a second language: trends in research and practice. Washington, D.C.: TESOL.
Brumfit, Christopher -- Rosamond Mitchell
1990 Research in the language classroom. London: Modem English Publications.
Burt, Marina K. -- Heidi C. Dulay (eds.)
1975 ON TESOL '75: new directions in second language learning, teaching and bilingual education. Washington, D.C.: TESOL.
Burt, Marina K. -- Carol Kiparsky
1972 The Gooficon: a repair manual for English. Rowley, Mass.: Newbury House.
Carroll, Susanne -- Y. Roberge -- Merrill Swain
1992 "The role of feedback in adult second language acquisition, error correction, and morphological generalizations", Applied Psycholinguistics 13: 173-198.
Cathcart, Ruth -- Judy Olsen
1976 "Teachers' and students' preferences for correction of classroom errors", in: John Fanselow -- Ruth Crymes (eds.), 41-53.
1986 "Teachers' priorities in correcting learners' errors in French immersion classes", in: Richard Day (ed.), 64-84.
Chenoweth, N. Ann. -- Richard Day -- Ann Chun -- Stuart Luppescu.
1983 "Attitudes and preferences of nonnative speakers to corrective feedback", Studies in Second Language Acquisition 6: 79-87.
Corder, S. Pit
1967 "The significance of learners' errors", International Review of Applied Linguistics 5: 161-170
1983 "Error evaluation: the importance of viewpoint", English Language Teaching Journal 37: 304-11.
Day, Richard (ed.)
1986 Talking to learn. Rowley, Mass.: Newbury House.
De Bot, Kees. -- Ralph B. Ginsberg -- Claire Kramsch (eds.)
1991 Foreign language research in cross-cultural perspective. Amsterdam: John Benjamins.
Doughty, Catherine -- Jessica Williams
1998 "Pedagogical choices in focus on form", in: Doughty, Catherine -- Jessica Williams (eds.), 197-261.
Doughty, Catherine -- Jessica Williams (eds.)
1998 Focus on form in classroom second language acquisition. Cambridge: CUP.
1985 "Discourse worlds in the classroom and in foreign language", Studies in Second Language Acquisition 7: 159-168.
1986 "Some ways in which teachers bring errors into being", in: Gabriele Kasper (ed.), 111-124.
1985 "Sources of variability in interlanguage", Applied Linguistics 6: 118-131.
1992 "The classroom context: an acquisition-rich or an acquisition-poor environment?", in: Claire Kramsch -- Sally McConnell-Ginet (eds.), 171-186.
Fanselow, John R. -- Ruth Crymes (eds.)
1977 On TESOL '76. Washington D.C.: TESOL.
1980 "Correction in native speaker-nonnative speaker conversation", in: Diane Larsen-Freeman (ed.), 125-137.
Gass, Susan M. -- Carolyn G. Madden (eds.)
1985 Input and second language acquisition. Rowley, Mass.: Newbury House.
Harley, Bridget -- Merrill Swain
1978 "An analysis of the verb system by young learners of French", Interlanguage Studies Bulletin 3: 35-79.
Hughes, Arthur -- Chryssoula Lascaratou
1982 "Competing criteria for error gravity", English Language Teaching Journal 36: 175-82.
1998 Errors in language learning and use: exploring error analysis. London: Longman.
Johnson, Karen E.
1996 Understanding Communication in the Second Language Classrooms. Cambridge: CUP.
1986 "Repair in foreign language teaching", in: Gabriele Kasper (ed.), 23-41.
Kasper, Gabriele (ed.)
1986 Learning, teaching and communication in the foreign language classroom. Aarhus: Aarhus University Press.
1985 "Communicative error evaluations: native speakers' evaluation and interpretation of written errors of Arab EFL learners", TESOL Quarterly 19: 225-351.
Kramsch, Claire -- Sally McConnel-Ginet (eds.)
1992 Text and context: cross-disciplinary perspectives on language study. New York: D. C. Heath.
Krashen, Stephen D.
1982 Principles and practice in second language acquisition. London: Pergamon.
Larsen-Freeman, Diane (ed.)
1980 Discourse analysis in second language research. Rowley, Mass.: Newbury House.
Lightbown, Patsy M.
1992 "Getting quality input in the second/foreign language classroom", in: Claire Kramsch -- Sally McConnell-Ginet (eds.), 187-197.
Lightbown, Patsy M. -- Nina Spada
1990 "Focus-on-form and corrective feedback in communicative language teaching: effects on second language learning", Studies in Second Language Acquisition 12: 429-48.
Long, Michael H.
1977 "Teacher feedback on learner error: mapping cognitions", in: H. Douglas Brown -- Carlos A. Yorio -- Ruth Crymes (eds.), 278-293.
1991 "Focus on form: A design feature in language teaching methodology", in: Kees de Bot -- Ralph B. Ginsberg -- Claire Kramseh (eds.), 39-52.
Long, Michael H. -- Robinson, Peter
1998 "Focus on form: Theory, research and practice", in: Catherine Doughty -- Jessica Williams (eds.), 15-41.
1992 Responses to correction of spoken errors in beginner and advanced adult learners. [Unpublished M.Ed. (TESOL) dissertation, University of South Australia.]
Nystrom, Nancy J.
1983 "Teacher-student interaction in bilingual classrooms: four approaches to error feedback", in: Herbert W. Seliger -- Michael H. Long (eds.), 169-188.
1984 "Psychological constraints on the teachability of languages", Studies in Second Language Acquisition 6: 186-214.
1989 "Is language teachable? Psycholinguistic experiments and hypotheses", Applied Linguistics 10: 52-79.
Ramirez, Arnulfo G. -- Nelly P. Stromquist
1979 "ESL methodology and student language learning in bilingual elementary schools", TESOL Quarterly 13: 145-158.
Richards, Jack C. -- Charles Lockhart
1994 Reflective teaching in second language classrooms. Cambridge: CUP.
1988 "Second language acquisition and its relationship to Universal Grammar", Applied Linguistics 9: 219-235.
Schegloff, Emmanuel A. -- Gail Jefferson -- Harvey Sacks
1977 "The preference for self-correction in the organization of repair in conversation", Language 53: 361-382.
Schmidt, Richard W.
1990 "The role of consciousness in second language learning", Applied Linguistics 11: 17-46.
1992 "Psychological mechanisms underlying second language fluency", Studies in Second Language Acquisition 14: 257-385.
Schmidt, Richard W. and Sylvia Frota
1986 "Developing basic conversational ability in a second language", in: Richard Day (ed.), 237-326.
Seliger, Herbert W. -- Michael H. Long (eds.).
1983 Classroom oriented research in second language acquisition. Rowley, Mass.: Newbury House.
1972 "Interlanguage", International Review of Applied Linguistics 10: 209-231.
1980 "Consciousness-raising and the second language learner", Applied Linguistics 2: 159-68.
1991 "Speaking to many minds: on the relevance of different types of language information for the L2 learner", Second Language Research 7: 118-132.
1993 "Input enhancement in instructed SLA: theoretical basis", Studies in Second Language Acquisition 15: 165-179.
1986 "Error perceptions of native speaking and non-native speaking teachers of ESL", English Language Teaching Journal 40: 306-12.
1985 "Communicative competence: some roles of comprehensible input and comprehensible output in its development", in: Susan Gass -- Carolyn G. Madden (eds.), 235-253.
1989 "Manipulating and complementing content teaching to maximize second language learning", TESL Canada Journal 6: 68-83.
1980 "Grammatical errors and communication breakdowns", TESOL Quarterly 14: 71-9.
1988 The classroom and the language learner. London: Longman.
Van Lier, Leo
1991 "Inside the classroom: learning processes and teaching procedures", Applied Language Learning 2. 29-69.
Vigil, Neddy A. -- John Oller
1976 "Rule fossilization: a tentative model", Language Learning 26: 281-295.
1987 "Against comprehensible input: the input hypothesis and the development of second language competence", Applied Linguistics 8: 95-110.
1991 "Adverb placement in second language acquisition. Some effects of positive and negative evidence in the classroom", Second Language Research 7: 122-161.